[Qemu-devel] Problem Running QEMU, and passing new commands

[Qemu-devel] Problem Running QEMU, and passing new commands

[Shane Brennan]

[-- Attachment #1: Type: text/plain, Size: 1366 bytes --]

Hi,
    I have successfully integrated the statistical simulation package
SimPoint with QEMU. Now I wish to add some commands to QEMU so a user can
start or stop simulation. The goal is that at the QEMU command prompt, the
user could enter a command like start_simpoint or stop_simpoint. For
example:

(qemu) start_simpoint
SimPoint started!
(qemu)
.... /*Time passes.. */
(qemu) stop_simpoint
SimPoint stopped!

I am curious where I would edit QEMU to look for such commands, some point
in the code where I can see what command is passed to QEMU, and do something
such as
"if strcmp(command, "start_simpoint") == 0"

Also, there is another problem, I am able to run QEMU if I tell it to load
and boot a linux image. But I am not able to simply boot up to the (qemu)
prompt and enter commands from there. I try running QEMU using the following
commands:
qemu-img create -f qcow c.img 3G
./i386-softmmu/qemu -hda c.img -L ./pc-bios/ -nographic

and it loads up to the (qemu) prompt. However, at that point everything
freezes. I am running QEMU on a linux host with a dual 64-bit AMD
processors, and doing this over an SSH connection.

If anyone has any advice on how I can get QEMU to boot to the point where I
can enter commands, and how I can monitor these commands within the source
code, I would very much appreciate the assistance.

~Shane Brennan
UC Santa Cruz

[-- Attachment #2: Type: text/html, Size: 1477 bytes --]

Re: [Qemu-devel] Problem Running QEMU, and passing new commands

[maestro]

> Hi,
>     I have successfully integrated the statistical simulation package
> SimPoint with QEMU. 
<no clue what simpoint is>
<snip>

> I am curious where I would edit QEMU to look for such commands, some
> point in the code where I can see what command is passed to QEMU, and
> do something such as 
> "if strcmp(command, "start_simpoint") == 0"
take a look a the monitor.c file in the root directory. 
"register" your function with the static term_cmd_t term_cmds[] and take
a look at the other commands for an example.

> qemu-img create -f qcow c.img 3G
> ./i386-softmmu/qemu -hda c.img -L ./pc-bios/ -nographic
with this you just create an empty disk and try to boot it. - i guess
this is similar to trying to boot an empty disk in a real system. the
-nographic switch prevents you from seeing the "unknown boot device"
error that is presented to you by the bios.
so do as with real hardware and install an operating system on that
c.img you just created (i'm fairly sure that is mentioned in the good
documentation provided on the qemu homepage ;-))

> and it loads up to the (qemu) prompt. However, at that point
> everything freezes. I am running QEMU on a linux host with a dual
> 64-bit AMD processors, and doing this over an SSH connection. 
you might try the -vnc option, you can then don't need -nographic nomore
and can check the system via any common vncclient.

> 
> If anyone has any advice on how I can get QEMU to boot to the point
> where I can enter commands, and how I can monitor these commands
> within the source code, I would very much appreciate the assistance.
i hope the above mentioned helped at least in parts and is not
completely wrong. if others think so please corret me.

> ~Shane Brennan 
> UC Santa Cruz
cheers
m.